Prosaposin Is Cleaved Into Saposins by Multiple Cathepsins in a Progranulin‐Regulated Fashion

ABSTRACT

Prosaposin (PSAP) is a lysosomal protein that plays a key role in sphingolipid metabolism. PSAP is cleaved into four bioactive disulfide‐rich saposins (SapA, SapB, SapC, and SapD) that catalyze sphingolipidases to promote sphingolipid breakdown. Maintaining optimal levels of PSAP and saposins is crucial for proper lysosomal function and sphingolipid homeostasis, and PSAP dysfunction is associated with juvenile‐onset lysosomal storage disorders and age‐associated neurodegenerative disorders. Despite this, the mechanism by which saposins are released from PSAP, and thus available to modulate sphingolipidases, sphingolipid homeostasis, and downstream lysosomal function, is not well understood. Here, we performed a comprehensive study to identify lysosomal enzymes that regulated prosaposin cleavage into saposins. In vitro cleavage assays identified multiple enzymes that could process human prosaposin into multi‐ and single‐saposin fragments. We confirmed the role of cathepsins D and B in PSAP processing and identified several additional lysosomal proteases (cathepsins E, K, L, S, V, G, and asparagine‐specific endopeptidase) that were able to process PSAP in distinctive, pH‐dependent manners. In addition, we found that PGRN and multi‐granulin fragments (MGFs) directly regulated the cleavage of PSAP by cathepsin D. With this study, we have shown that multiple cathepsins, PGRN, and MGFs work in concert to produce saposins under different conditions, which could present novel opportunities to modulate saposin levels in disease.

Prosaposin (PSAP) is a lysosomal protein cleaved into four bioactive saposins (SapA‐D) that regulate sphingolipid breakdown. Here, we identify nine cathepsins, including seven newly implicated enzymes, that process PSAP in a pH‐dependent manner to generate distinct cleavage products. We further show that the multi‐granulin fragment CDE, derived from progranulin (PGRN), directly enhances PSAP cleavage by cathepsin D. Using human neurons, CRISPR interference demonstrates that cathepsins D, L, and AEP are critical for PSAP processing, highlighting how multiple proteases and regulators coordinate saposin production to maintain lysosomal function and sphingolipid homeostasis. Overall, our study shows that cathepsins and multi‐granulin fragments act together to generate saposins under diverse conditions, revealing potential avenues to modulate saposin levels and downstream sphingolipid homeostasis.

Abbreviations

AEP/LGMN

Asparagine‐specific endopeptidase

CRISPRi

CRISPR interference

CTSA

Cathepsin A

CTSB

Cathepsin B

CTSC

Cathepsin C

CTSD

Cathepsin D

CTSE

Cathepsin E

CTSF

Cathepsin F

CTSG

Cathepsin G

CTSH

Cathepsin H

CTSK

Cathepsin K

CTSL

Cathepsin L

CTSO

Cathepsin O

CTSS

Cathepsin S

CTSV

Cathepsin V

CTSW

Cathepsin W

CTSX

Cathepsin X

GrnD

Granulin D

HLL

human liver lysosome

ICLAC

International Cell Line Authentication Committee

iPSC

induced pluripotent stem cell

MGF

multi‐granulin fragment

NTC

non‐targeting control

PepA

Pepstatin A

PGRN

Progranulin

PSAP

Prosaposin

PTM

Post‐translational modification

RRID

Research Resource Identifier (see scicrunch.org)

SapA

Saposin A

SapB

Saposin B

SapC

Saposin C

SapD

Saposin D

1. Introduction

A key factor in neurodegenerative disease is the progressive deterioration of protein and lipid homeostasis, leading to accumulation of insoluble macromolecules and cellular dysfunction. Given their role in the degradation and recycling of macromolecules, lysosomes have emerged as important contributors to cellular homeostasis. Moreover, lysosomal dysfunction is linked to juvenile‐onset lysosomal storage and adult‐onset neurodegenerative diseases (Nixon 2020). Specifically, perturbed lysosomal metabolism of sphingolipids also represents a risk factor for the onset and progression of neurodegenerative diseases (Park et al. 2020; Di Pardo and Maglione 2018; Johnson and Stolzing 2019; Castellanos et al. 2021; Schulze and Sandhoff 2014; Hernandez‐Diaz and Soukup 2020).

Sphingolipid metabolism in the lysosome is driven by sphingolipidases. Sphingolipidase activator proteins, also known as SAPs or saposins, are bioactive disulfide‐rich peptides that catalyze sphingolipidase activity. Saposins regulate this process both through direct activation of sphingolipidases and by disrupting the intralysosomal membrane to present substrates to these enzymes (Vaccaro et al. 1999; Rossmann et al. 2008; Qi and Grabowski 2001; Remmel et al. 2007). Prosaposin (PSAP), the evolutionarily conserved precursor for the four major SAPs, SapA, SapB, SapC, and SapD (Figure 1A) (Darmoise et al. 2010; Kishimoto et al. 1992), is central to the sphingolipid catabolism pathway (Darmoise et al. 2010; Morimoto et al. 1989; Morimoto et al. 1990; Matsuda et al. 2001; Kolter and Sandhoff 2006; Vogel et al. 1991; Ho and O'Brien 1971; Berent and Radin 1981; Wenger et al. 1982; Tayama et al. 1993; Azuma et al. 1994; Matsuda et al. 2004; Klein et al. 1994). Indeed, mice lacking PSAP accumulate several sphingolipids (Darmoise et al. 2010; Bradova et al. 1993; Fujita et al. 1996). Further, loss of PSAP in humans leads to Combined Saposin Deficiency, a fatal infantile lysosomal storage disorder with severe neurological pathology (Hulkova 2001). Loss of each individual saposin leads to a different lysosomal storage disorder. Deficiency in SapA, SapB, or SapC leads to Krabbe disease (Kolter and Sandhoff 2006; Spiegel et al. 2005; Won et al. 2013; Kose et al. 2019), metachromatic leukodystrophy (Darmoise et al. 2010; Li et al. 1985; Wenger et al. 1989; Kretz et al. 1990; Schlote et al. 1991; Sun et al. 2008), or Gaucher disease (Darmoise et al. 2010; Matsuda et al. 2004; Christomanou et al. 1989; Schnabel et al. 1991), respectively.

FIGURE 1.

Full‐length prosaposin is proteolyzed by multiple lysosomal proteases in vitro. (A) Model of prosaposin (PSAP) cleavage into saposins. (B) Human recombinant PSAP (400 nM) was incubated at 37°C for a 6 h time course at pH 4.5 with human liver lysosomes (HLLs, 50 ng/μL) with and without the aspartyl protease inhibitor Pepstatin A (PepA, 5 μM). Samples were run on western blot and probed with polyclonal α‐PSAP. (C–E) PSAP (400 nM) was incubated at 37°C for 1 h with human recombinant cathepsins (400 nM) under four pH conditions as indicated. Cleavage fragments were visualized by silver stain. *Indicates the anticipated molecular weight of each cathepsin. Data represent three independent replicates. For a summary of cleavage capacity, see Table S1.

In addition to sphingolipidases, PSAP directly interacts and traffics with another lysosomal protein, progranulin (PGRN) (Zhou, Sullivan, et al. 2017; Zhou, Sun, et al. 2017). PGRN loss of function mutations also lead to neurodegenerative diseases such as Frontotemporal Dementia (FTD) (Baker et al. 2006; Cruts et al. 2006). Recently, PGRN has been implicated in disease‐related dysfunction in sphingolipid homeostasis (Evers et al. 2017; Huang et al. 2020). In the lysosome, PGRN is cleaved by a subset of lysosomal proteases sequentially into multi‐granulin fragment (MGF) intermediates and ultimately bioactive granulins (Mohan et al. 2021). There, PGRN and the MGFs promote the activation of CTSD (Butler, Cortopassi, Argouarch, et al. 2019; Butler, Cortopassi, Gururaj, et al. 2019; Zhou, Paushter, et al. 2017; Tayebi et al. 2020; Kao et al. 2017). As CTSD has been shown to catalyze the cleavage of PSAP into saposins (Hiraiwa et al. 1997), this is one potential mechanism by which PGRN can indirectly influence lipid metabolism.

Despite the important roles for saposins A‐D in sphingolipid metabolism and disease, the mechanism by which saposins are released from PSAP is poorly understood. As a lysosome resident protein, PSAP proteolysis likely occurs in this organelle (Tayebi et al. 2020; Hiraiwa et al. 1993). Indeed, lysosomal proteases CTSD (Hiraiwa et al. 1997) and cathepsin B (CTSB) (Kim et al. 2022), have been shown to cleave PSAP into saposins. However, a comprehensive study of the PSAP cleavage capacity of lysosomal proteases has not been performed. In this study, we set out to understand the processing and regulation of PSAP into saposins by lysosomal proteases. Using in vitro protease cleavage assays, we identified several lysosomal proteases that could process PSAP into trisaposins, disaposins, and saposins and show that this processing occurred in a protease‐specific and pH‐dependent manner. Additionally, we found that PGRN and MGFs have differing effects on PSAP cleavage, and these effects were mediated by CTSD.

2. Materials and Methods

2.1. In Vitro Cleavage Assays

For these assays, 400 nM of recombinant human PSAP (Abcam cat. no. #167924) was incubated with or without 50 ng/μL of human liver lysosomes (XenoTech cat. no. #H0610.L) or 400 nM of each protease and with or without 400 nM of recombinant human PGRN (Fisher cat. no. #2420‐PG) or recombinant human MGFs pG (Mybiosources cat. no. #2012256), BAC (Mybiosources cat. no. #2011118), or CDE (Mybiosources cat. no. #2018332). To inhibit aspartyl proteases, 5 μM of Pepstatin A (Millipore cat. no. #516485) was used. Reactions at pH 3.5 were performed in 100 mM sodium citrate, reactions at pH 4.5 and 5.5 were performed in 50 mM sodium acetate, and reactions at pH 7.4 were performed in 100 mM phosphate buffer saline (PBS). 20 μL reactions were incubated with 1 mM EDTA and 2 mM DTT for 60 min in a 37°C water bath in LoBind Tubes (Fisher cat. no. #E925000090). Protease activity was stopped by adding 7.5 μL of NuPAGE 4X LDS (Fisher cat. no. #NP0007), 3 μL of 10X reducing agent (Fisher cat. no. #NP0009), and denatured for 10 min at 80°C.

2.2. Recombinant Proteases

The following cathepsins were used in this study: CTSA (R&D cat. no. #1049‐SE), CTSB (Millipore cat. no. #219364), CTSC (R&D cat. no. #1071‐CY), CTSD (Sigma cat. no. #C8696), CTSE (R&D cat. no. #1294‐AS), CTSF (Abcam cat. no. #157039), CTSG (Millipore cat. no. #219873), CTSH (R&D cat. no. #7516‐CY‐010), CTSK (Millipore cat. no. #219461), CTSL (Millipore cat. no. #219402), CTSO (Abcam cat. no. #ab158237), CTSS (R&D cat. no. #1183‐CY), CTSV (R&D cat. no. #1080‐CY), CTSW (Abcam cat. no. #ab158238), CTSX (R&D cat. no. #934‐CY), Asparagine Endopeptidase/Legumain (R&D cat. no. #2199‐CY). In addition to purity testing performed by the manufacturers, our lab has independently verified the purity of these recombinant proteases (Sampognaro et al. 2023).

2.3. Activation of Proteases

For CTSA and CTSC activation, 4 μM of each protease was incubated with 200 nM of Cathepsin L at room temperature for 60 min. After 60 min, 50 μM of benzyloxycarbonyl FY (t‐Bu)‐DMK (Sigma cat. no. #219427), an irreversible, highly specific inhibitor of CTSL was added to quench activity. For CTSH activation, 4 μM of CTSH was incubated with 500 nM of thermolysin (R&D cat. no. #3097‐ZN) at room temperature for 3 h. After 3 h, 1 mM of Phosphoramidon (Tocris Bioscience cat. no. #6333), a specific Thermolysin inhibitor, was added to quench Thermolysin activity. All other recombinant proteases are active, as shown by the vendors and previously in our lab (Sampognaro et al. 2023).

2.4. Silver Staining

Silver staining was performed according to the manufacturer's instructions with SilverQuest silver staining kit (Thermo cat. no. #LC6070).

2.5. SDS‐PAGE and Western Blotting

Samples were resolved on 4%–12% polyacrylamide gels (Invitrogen, cat. no. #NP0322) using NuPAGE MOPS SDS running buffer (Invitrogen, cat. no. #NP0001) at 150 V for approximately 1 h. Proteins were transferred to nitrocellulose membranes using the wet transfer method with NuPAGE transfer buffer (Invitrogen, cat. no. #NP00061) at 300 mA for 1 h. Following transfer, membranes were blocked with Odyssey buffer (Li‐cor, cat. no. #927‐50 010) for 1–2 h at room temperature. Membranes were incubated overnight at 4°C with primary antibodies diluted 1:1000 in Odyssey blocking buffer. The following primary antibodies were used: α‐PSAP pAb (Abcam cat. no. #180751), α‐SapA pAb (Proteintech cat. no. #18396‐1‐AP, RRID:AB_2172460), α‐SapB (Proteintech cat. no. #18397‐1‐AP, RRID:AB_10598628), α‐SapC (Proteintech cat. no. #18398‐1‐AP, RRID:AB_10598315), and/or α‐SapD (Proteintech cat. no. #18423‐1‐AP, RRID:AB_10643525). Membranes were washed three times for 10 min each with TBST (Tris‐buffered saline with 0.1% Tween‐20) and incubated with Li‐Cor fluorescent secondary antibodies at 1:10000 dilution for 1 h at room temperature. After washing the membranes three more times with TBST, the blots were imaged using the Odyssey CLx imaging system (Li‐Cor).

2.6. Cathepsin D Activity Assay

Cathepsin D activity assays were performed according to the manufacturer's instructions using a Fluorometric Cathepsin D Activity Assay Kit (Ab cat. no. #65302).

2.7. CRISPR‐Modulated iNeurons

KOLF2.1J iPSCs were previously obtained and commercially validated through the Jackson Laboratory (in collaboration with the iPSC Neurodegenerative Disease Initiative [iNDI]). The cell line was last authenticated in 2023 by the Jackson Laboratory (JAX) using their GDA‐Cyto assay for validation and are not listed as misidentified by ICLAC. These cells were used at a maximum passage number of 18. iPSCs were transfected with a doxycycline‐inducible piggyBac system carrying the NGN2 gene. After stable transfection with inducible NGN2, these lines subsequently underwent lentiviral infection with a CRISPRi all‐in‐one dCas9‐SALL1‐SDS3 + sgRNA for either CTSD, CTSL, or AEP inhibition. These CRISPRi constructs were commercially obtained via Dharmacon. Successful knockdown of each lysosomal protease was subsequently confirmed by qtPCR.

2.8. iNeuron Differentiation With NGN2

On Day 0, iPSCs were dissociated with Accutase, quenched with DPBS, and then resuspended in StemFlex with 10 μM ROCK inhibitor. Cells were counted and plated at appropriate densities (4 million cells in a 10‐cm dish or 0.66 million cells in a well of a 6‐well plate) in Neural Induction Media (NIM) (DMEM/F12, N2 and NEAA) supplemented with 2 μg/mL doxycycline, 1:2000 dilution of BDNF, 1:2000 dilution of NT3, 1 μg/mL laminin and 10 μM ROCK inhibitor. On Day 1, the medium was replaced with fresh NIM of the same composition as Day 0, but without ROCK inhibitor. On Day 3, cells were replated and resuspended in fresh Neural Maturation Media (NMM) (DMEM/F12, Neurobasal‐A, N2, B27, NEAA, GlutaMAX) supplemented with 2 μg/mL doxycycline, 1:2000 dilution of BDNF, 1:2000 dilution of NT3, and 1 μg/mL laminin. Cells were counted and plated at the desired density (500 000 cells/well in a 6‐well plate) on poly‐d‐lysine (PDL)‐coated plates. From Day 7 onward, cells were fed with media changes every 3 days, with the media being composed of NMM supplemented with 1:2000 dilution of BDNF, 1:2000 dilution of NT3, and 1 μg/mL laminin.

2.9. Generation and Use of SH‐SY5Y LysoIP Line

SH‐SY5Y human neuroblastoma cells were obtained from American Type Culture Collection (ATCC, CRL‐2266). The cell line was last authenticated in 2023 by ATCC using their ATCC authenticates cell lines using Short Tandem Repeat (STR) profiling for validation and are not listed as misidentified by ICLAC. These cells were used at a maximum passage number of 16. Cells were maintained with 1:1 EMEM/F12 media (ATCC cat. no. #30‐2003/Thermo cat. no. #11765062) supplemented with 10% FBS and 1% Penicillin–Streptomycin (PS) (Thermo Fisher cat. no. #15140122). Lentiviruses were produced by transfecting HEK‐293FT cells with pLJC5‐Tmem192‐3xHA construct obtained from Addgene (RRID:Addgene_102930) in combination with VSV‐G and CMV‐ΔVPR packaging plasmids (Abu‐Remaileh et al. 2017). The following morning, the media was refreshed. Three days after transfection, the virus‐containing supernatant was collected, then centrifuged at 1000×g to remove debris and then frozen at −80°C. Virus was concentrated using Lenti‐X. To establish stably expressing cell lines, 300 000 SH‐SY5Y cells were plated in 6‐well plates in F12/EMEM with 10% FBS and PS and 8 μg/mL polybrene and infected with 5 μL of virus‐containing media. The following morning, the media was refreshed and puromycin was added for selection.

LysoIP was performed as previously described (Abu‐Remaileh et al. 2017). In brief, 1.5e6 LysoIP SH‐SY5Y cells were plated in 15 cm plates. To terminally differentiate into neuron‐like cells, SH‐SY5Y cells were then treated with retinoic acid as previously published (Sampognaro et al. 2023). On day 10, the cells were rinsed with PBS, then collected in 5 mL PBS. The cells were spun at 1000 g × 2 min. The pellet was resuspended in 1000 μL PBS. The suspension was homogenized in a dounce homogenizer for 20 strokes, then spun at 1000 g × 2 min. The supernatant was incubated with pre‐washed anti‐HA beads for 15 min on a rotator in the cold room. The beads were washed 3 times with PBS. The lysosomes were eluted by incubation with 0.5% NP40 in PBS for 10 min on rotator in the cold room. Total protein concentration was measured and 12.5 ng/μL of lysosomes was used for each reaction in place of recombinant cathepsin, as described above.

2.10. Statistical Analyses

All data were quantified from Western blot experiments and are presented as mean ± SEM from three independent biological replicates. Statistical analyses were performed using GraphPad Prism Version 10.5.0. No a priori statistical methods were used to predetermine sample size. The number of replicates (n = 3) for Western blots and qPCR was chosen based on standard practice in the field and prior studies assessing protein abundance and CRISPRi knock‐down. Due to the small sample size, formal tests of normality and outliers were not performed. For CRISPRi qPCR fold‐change data, each knockdown line was compared to the baseline value of 1 using two‐tailed one‐sample t‐tests. For Western blot data, comparisons between knockdown conditions and the non‐targeting control (NTC) were performed using one‐way ANOVA followed by Dunnett's multiple‐comparisons test. All tests were two‐tailed, and statistical significance was defined as p < 0.05. Exact F statistics, degrees of freedom, and adjusted p‐values for each comparison are reported in the figure legends.

3. Results

3.1. Full‐Length Prosaposin Is Proteolyzed by Multiple Lysosomal Proteases In Vitro

To determine if full‐length PSAP can be processed in the lysosome, we incubated recombinant PSAP with isolated human liver lysosomes (HLLs) in vitro, and cleavage products were visualized via western blot (Figure 1B). Over a six‐hour incubation at 37°C at a lysosomal pH of 4.5, HLLs cleaved PSAP into fragments that were approximately molecular weight of di‐ (~35 kDa) and tri‐saposins (~45 kDa). HLLs consist of numerous proteolytic enzymes, including glycosidases, which may be responsible for the downward shift in the multi‐saposin bands. HLLs contain the aspartyl protease CTSD (Figure S1), which has a well‐defined role in PSAP processing (Hiraiwa et al. 1997). To determine if CTSD is the primary driver of PSAP cleavage by HLLs, we treated them with the aspartyl protease inhibitor Pepstatin A (PepA) to block CTSD function. We found that while PepA treatment inhibited CTSD activity in HLLs (Figure S1B), it had little to no impact on the ability of HLLs to cleave PSAP, suggesting that additional proteases found in HLLs are capable of processing PSAP.

To identify the lysosomal proteases that cleave PSAP, we performed individual in vitro assays with commercially available recombinant human lysosomal cathepsins (CTS). In addition to purity testing performed by the manufacturers, our lab independently verified the purity and activity of these recombinant proteases (Sampognaro et al. 2023). Recombinant PSAP was incubated with individual enzymes for 1 h at 37°C and visualized via silver stain. As lysosomal proteases have distinct pH preferences that can be substrate‐dependent (Turk et al. 2012, 1999; Banay‐Schwartz et al. 1985), we performed the study across a range of pH settings (pH 3.5, 4.5, 5.5, and 7.4). We first assayed the aspartyl proteases, cathepsins D and E. Aligning with previous reports, we found that CTSD hydrolyzed PSAP into discrete peptides (Figure 1C). However, despite being one of the only known cathepsins for PSAP (Hiraiwa et al. 1997), mature CTSD did not induce strong cleavage, even at its optimal pH of 3.5. On the other hand, we found that the second aspartyl protease CTSE almost fully reduced PSAP to saposins at pH 3.5 and 4.5 (Figure 1C). CTSD and CTSE are highly related and recognize similar amino acid motifs (Yasuda et al. 1999). Nonetheless, they exhibit different abilities to process PSAP into saposins, and there are subtle differences between preferred cleavage sites that may explain the pronounced differences in activity seen here.

Given that HLLs were still able to proteolyze PSAP in the presence of PepA, we assessed serine and cysteine proteases for their ability to cleave PSAP. Their activity can be categorized into three groups: non‐cleavers, saposin‐producers, and those that fully digested PSAP. Non‐cleavers, which included CTSA, H, O, and W, did not digest PSAP at any pH tested (Figure 1D,E). Saposin producers, including CTSG, B, K, L, S, V, and asparagine‐specific endopeptidase (AEP), generated distinct cleavage profiles, producing fragments of mono‐, di‐, and tri‐saposins (Figure 1D,E). Of these, all but CTSG had optimal activity at an acidic pH. Our finding that CTSB proteolyzes PSAP is consistent with previous research (Kim et al. 2022). Notably, cleavage by CTSB appeared to preferentially occur between SapB and SapC, as the major product was the size of disaposins. While CTSG, S and AEP produced saposins, CTSL and CTSV were less efficient and did not produce discernable saposins at the time point tested. Finally, CTSK was unique in that it produced saposins and disaposins at pH 3.5 and 7.4 but completely digested PSAP at pH 4.5 and 5.5. Similarly, cathepsins F (at pH 4.5), C and X (at pH 3.5), digested some PSAP without producing saposins (Figure 1D,E).

In total, two aspartyl proteases (CTSD and E), one serine protease (CTSG), and six cysteine proteases (CTSB, K, L, S, V, and AEP) produced saposins, and three cysteine proteases (CTSF, C, and X) digested full‐length PSAP at variable pH settings when incubated individually (Table S1). These in vitro protease assays also highlighted the variable kinetics of lysosomal protease processing of PSAP. Over 1 h, some proteases (e.g., CTSK at pH 4.5 and 5.5) completely degraded PSAP, while others (e.g., CTSG and CTSS) produced different‐sized, intermediate bands depending on pH. The step‐wise sized bands suggested that mono‐, di‐, and tri‐saposin fragments were being produced. Together, these results suggested that PSAP processing is likely dependent on several different cathepsins in vivo. Since protease expression can be cell‐type specific (Mohan et al. 2021), these results suggested that PSAP processing may differ between cell types.

To further understand the diversity of PSAP cleavage products produced by each cathepsin, we used the PROSPER database to predict intersaposin cleavage sites in silico (Figure 2, Table S2) (Song et al. 2012). All cathepsins available in the database at the time of analysis were queried. CTSD, E, B, K, and L were predicted to cleave in all intersaposin regions. PROSPER analysis only predicted CTSS to cleave between SapA‐SapB and SapC‐SapD, and CTSG to cleave in the intersaposin region between SapA‐SapB and SapB‐SapC. We also considered that disease mutations and post‐translational modifications (PTMs) (Table S3) might affect PSAP cleavage (Figure 2). Of the known disease mutations, none fall in intersaposin regions. Nineteen are within saposin domains, two are at the start codon, and two occur before SapA. On the other hand, there are several PTMs, particularly ubiquitination, phosphorylation, and O‐GalNAclyation, which occur in predicted intersaposin cleavage sites which may disrupt proteolysis. These PTMs may fine‐tune PSAP processing or stability in a context‐dependent manner, and exploring how they influence cathepsin‐mediated cleavage will be an interesting direction for future studies.

FIGURE 2.

A string of pearls model shows several prosaposin predicted cleavage sites at intersaposin regions, while mutation sites cluster in saposin domains. Cathepsin cleavage sites were predicted in silico using the PROSPER database (Mittenbuhler et al. 2023). All cathepsins available in the database were queried. For a list of predicted cleavage sites, see Table S2. Post‐translational modifications (PTMs) and disease‐causing mutations were identified via literature review. For a list of predicted PTM and mutation sites, see Table S3.

3.2. Cathepsins Produce Saposins and Multi‐Saposin Fragments at Different Rates In Vitro

To better resolve the stepwise proteolysis of PSAP into saposins, the nine cathepsins that appeared to cleave PSAP into discrete fragments at 1 h were tested at their optimal pH setting over a time course. The degree to which they digested PSAP in 1 h differed significantly. For example, CTSE, G, B, K, and S completely digested full‐length PSAP at 1 h at multiple pHs. In contrast, while CTSD, L, V, and AEP were PSAP cleavage competent, they did so at a slower rate. Thus, we chose to test the ability of these proteases to produce saposins in two groups. “Efficient” proteases were tested over the course of 30 min, and “inefficient” proteases were tested over 6 h.

The group of “efficient” proteases, CTSE, G, B, K, and S, all produced all four saposins by 30 min, albeit with different relative efficiencies. The serine protease CTSG was the most effective protease tested and completely digested PSAP into saposins after a 5‐min incubation (Figure 3E). CTSK, E, and B also produced all saposins after a 5‐min incubation (Figure 3A–C), however, they took up to 30 min to completely digest PSAP. CTSS produced only low levels of all saposins by 30 min (Figure 3D). Of the “inefficient” proteases, CTSD and AEP also produced all four saposins, albeit after a much longer time course (Figure 3F–I). Of those tested over the longer time course, AEP was the most efficient and completely digested PSAP by 3 h. CTSD was able to produce all four saposins by 3 h. Finally, CTSL was able to produce SapA and SapD (Figure 3H), and CTSV was able to produce SapD (Figure 3F), but no appreciable quantities of the other saposins by 6 h.

FIGURE 3.

Cathepsins produce saposins (A–D) and multi‐saposin fragments at different rates in vitro. Human recombinant prosaposin (PSAP, 400 nM) was incubated at 37°C for a time course depending on their rate of cleavage. (A–E) CTSE, B, K, S, and G (400 nM) cleave PSAP were tested over 30 min, while (F–I) CTSD, L, V, and AEP (400 nM) were tested over 6 h. Samples were run on western blot and probed with α‐SapA, α‐SapB, α‐SapC, α‐SapD antibodies and total PSAP, SapA, SapB, SapC, and SapD were quantified and normalized to PSAP at time zero on their respective blots, shown in Figure S2. All enzymes were tested at pH of 4.5, with the exception of CTSG, which was tested at its optimal pH of 7.4. Data represent three independent replicates.

These experiments demonstrated that most cathepsins that cleave PSAP were able to produce all four saposins with varying efficiency in vitro, consistent with in silico predictions. Amounts of PSAP and saposins A‐D at individual timepoints were quantified over three replicates (Figure S2). For all proteases tested, SapA (9–21 kDa) was the most abundant saposin present at the shortest incubation time (Figure 3), consistent with the previous observation that SapA is the first saposin released from PSAP (Hiraiwa et al. 1997). Accordingly, the remainder of the protein, Trisaposin BCD (45 kDa), was also present at high quantities at these time points. Less commonly, the first cleavage of PSAP also occurred between SapB and SapC, generating Disaposin AB (35 kDa) and Disaposin CD (35 kDa). Cathepsins rarely cleaved full‐length PSAP between SapC and SapD, as indicated by the rarity of Trisaposin ABC (48 kDa, see IB:SapA in Figure 3) and the relative delay of the presence of SapD (13 kDa), which must instead be digested from the downstream products Trisaposin BCD and Disaposin CD (Figure 3). Disaposin CD abundance persisted longer than Disaposin AB, and SapC (9 kDa) was often the last saposin to be produced. Together, these data indicate that certain saposins are more readily produced than others. Considering the unique role for each saposin in sphingolipid metabolism, the ability of various cathepsins to produce less abundant saposins, like SapC, may be critical to the maintenance of sphingolipid homeostasis.

3.3. Cathepsin Knockdown Impairs PSAP Cleavage and Saposin Production in iPSC‐Derived Neurons

To determine whether individual cathepsins are required for PSAP cleavage in neurons, we used CRISPR interference (CRISPRi) in human iPSC‐derived neurons. These neurons provide a physiologically relevant context to assess PSAP processing in a cell type where lysosomal function and lipid metabolism are highly active. We focused on CTSL, AEP, and CTSD, three lysosomal proteases that are highly expressed in the nervous system and have been implicated in neuronal proteostasis and neurodegeneration.

Our experimental workflow is summarized in Figure 4A, which illustrates the CRISPRi‐mediated knockdown strategy and neuronal differentiation. KOLF2.1J iPSCs were first transfected with doxycycline‐inducible piggyBac system carrying the NGN2 gene. After stable transfection with inducible NGN2. Next, these lines underwent lentiviral infection with a CRISPRi all‐in‐one dCas9‐SALL1‐SDS3 + sgRNA for either CTSL, AEP, or CTSD inhibition. Finally, cells were differentiated into neurons via NGN2 induction. Efficient suppression of CTSL, AEP, and CTSD expression was confirmed by quantitative PCR (Figure 4B). Western blot analysis revealed that knockdown of each protease led to an accumulation of full‐length PSAP (Figure 4C,D). Notably, we also observed a pronounced increase in a trisaposin band detected by the SapC antibody, suggesting incomplete processing of PSAP intermediates (Figure 4C,D). Although a reduction in SapC was visually apparent, its abundance was near the lower limit of detection for our western blot assays, making precise quantification difficult. Together, these data identify CTSL, AEP, and CTSD as major contributors to PSAP processing in neurons, suggesting that altered activity of these cathepsins could impact saposin availability and lysosomal lipid homeostasis relevant to neurodegenerative disease.

FIGURE 4.

Knockdown of lysosomal cathepsins disrupts PSAP processing in iPSC‐derived neurons. (A) Schematic of the CRISPRi modification and neuronal differentiation protocol used to generate iPSC‐derived neurons (Biorender). (B) Knockdown efficiency of CTSL, AEP, and CTSD was confirmed by qPCR. Fold‐change in CRISPRi lines was compared using two‐tailed one‐sample t‐tests (n = 3). CTSLi: T(2) = −15.26, p = 0.0042; AEPi: T(2) = −5.07, p = 0.037; CTSDi: T(2) = −6.35, p = 0.024. (C) Western blot analysis of PSAP and saposin C in control and cathepsin knockdown neurons probed with α‐PSAP, α‐SapC, and α‐GAPDH antibodies. (D) Quantification of PSAP and Trisaposin bands from western blots in shown in C. Data represent three independent replicates (mean ± SEM). For PSAP, one‐way ANOVA showed no significant effect (two‐tailed; F(3, 8) = 1.29, p = 0.34). For Trisaposin, one‐way ANOVA showed no significant effect (two‐tailed; F(3, 8) = 3.88, p = 0.056).

3.4. Multi‐Granulin Fragments Originating From Progranulin Increase Cathepsin D Processing of Prosaposin

The PSAP binding partner progranulin (PGRN) has been shown to regulate saposin abundance (Zhou, Sun, et al. 2017; Valdez et al. 2020) and it has been recently implicated in disease‐related dysfunction in lipid homeostasis (Evers et al. 2017; Huang et al. 2020). PGRN directly binds to and regulates the subcellular localization of PSAP (Zhou, Sullivan, et al. 2017; Zhou, Sun, et al. 2017). Further, PGRN and multi‐granulin fragments (MGFs) regulate the activity of CTSD (Butler, Cortopassi, Argouarch, et al. 2019; Butler, Cortopassi, Gururaj, et al. 2019; Zhou, Paushter, et al. 2017). Based on this, we hypothesized that PGRN may play a role in PSAP cleavage via modulation of CTSD activity.

Similar to PSAP, PGRN is a lysosomal pro‐protein that is cleaved into discrete MGFs and individual granulins by a number of cathepsins (Figure 5A) (Mohan et al. 2021). The GrnD domain of PGRN directly binds to the intersaposin region between SapB and SapC of PSAP (Zhou, Sullivan, et al. 2017; Zhou, Sun, et al. 2017). In addition to recombinant full‐length PGRN, we tested commercially available MGFs, pG, BAC, and CDE for their ability to regulate PSAP cleavage by CTSD. In addition to purity testing (Figure S3). To test the effect of PGRN and MGFs on CTSD‐mediated cleavage of PSAP, we incubated PSAP with CTSD and PGRN or MGFs and measured saposin production over a 6 h time course. Interestingly, incubation of CTSD with PGRN decreased PSAP cleavage in vitro (Figure 5B–E). While incubation of CTSD with pG did not change PSAP cleavage, incubation with BAC or CDE increased the rate of PSAP cleavage (Figure 5B–E). While CTSD alone completely digested PSAP after 6 h, co‐incubation of CTSD with CDE allowed for complete digestion of PSAP by only 1 h.

FIGURE 5.

Progranulin and multi‐granulin fragments influence the rate of PSAP cleavage and production of saposins (A–D). (A) Model of progranulin (PGRN) showing full length, and the three multi‐granulin fragments (MGFs) pG, BAC, and CDE. *Indicates the PSAP binding site on PGRN. (B–E) PSAP was incubated at 37°C for a 6 h time course with CTSD (400 nM) and full‐length PGRN, pG, BAC, or CDE (400 nM) at pH 4.5. (F–I) PSAP was incubated with CTSD and CDE as in (B–E) but a shorter time course of 30 min. Samples were run on western blot and probed with α‐SapA, α‐SapB, α‐SapC, α‐SapD antibodies. (J) Schematic of lysosome isolation from SH‐SY5Y neuronal culture by LysoIP (Biorender). (K) PSAP was incubated at 37°C for up to 6 h with SH‐SY5Y LysoIP lysosomes (12.5 ng/μL) with or without CDE (400 nM) at pH 4.5. Data represents three independent replicates.

Considering that the addition of CDE caused CTSD to completely digest PSAP into saposins at the shortest timepoint measured for the “inefficient” proteases, we sought to determine if CDE could enhance CTSD processing of PSAP to the level of the “efficient” proteases. To test this, we performed a shorter time course of 30 min (Figure 5F–I). Our results showed that incubation of CTSD with CDE facilitated saposin production within 5 min. Compared to all the proteases tested for PSAP processing, CTSD alone was one of the least efficient, and CTSD with CDE is among the most efficient. This suggests a novel mechanism for PGRN in the regulation of PSAP, positioning it as an upstream regulator of PSAP cleavage and saposin production.

Because cathepsin expression and regulation vary across tissues, we next asked whether the addition of CDE would influence PSAP cleavage in a neuronal context. To test this, we isolated neuronal lysosomes from SH‐SY5Y cells expressing a LysoIP tag and incubated them with recombinant PSAP in the presence or absence of CDE over a 6 h time course (Figure 5J). Consistent with our in vitro findings, the addition of CDE enhanced PSAP cleavage by neuronal lysosomes (Figure 5K). Together, these results suggest that PGRN‐derived fragments such as CDE can potentiate PSAP processing within neuronal lysosomes, highlighting a potential mechanism for neuronal regulation of saposin production.

4. Discussion

In this study, we surveyed lysosomal cathepsins to determine which proteases are capable of PSAP cleavage, the pH‐dependence of their activity, and their ability to produce specific saposin species. Considering the key role of saposins in sphingolipid metabolism and lysosomal storage disorders, a better understanding of the proteases responsible for saposin production has potential therapeutic implications. Previously, the only known proteases to cleave PSAP into saposins were CTSD (Zhou, Sullivan, et al. 2017) and CTSB (Zhou, Sun, et al. 2017). Here, we showed that these cathepsins and multiple others (CTSE, K, L, S, V, G and AEP) can process PSAP into saposins. Moreover, we found that the majority of proteases that process PSAP do so predominantly at a lysosomal pH of 3.5–4.5, suggesting that the majority of PSAP processing occurs in the lysosome, consistent with the canonical role of saposins in modulating lysosomal shingolipidase activity. Surprisingly, our data indicate that PSAP cleavage can also occur at a neutral pH 7.4 by cathepsins G, B, K, and S. All four of these cathepsins can be secreted (Turk et al. 1999; Song et al. 2012; Valdez et al. 2020; Meyer et al. 2014; Mittenbuhler et al. 2023), suggesting that they may also cleave PSAP extracellularly. While extracellular prosaposin exerts effects on G‐protein coupled receptor signaling and ERK phosphorylation (Meyer et al. 2014; Mittenbuhler et al. 2023), extracellular saposins are relatively understudied.

Our data suggests that many cathepsins can contribute to the generation of saposins A‐D, and thus may influence various aspects of sphingolipid metabolism and lysosomal storage disorders. These disorders result primarily from deficiencies in specific sphingolipid hydrolases, with saposins acting as essential cofactors for their activity. Indeed, research has shown that saposin‐linked lysosomal storage disorders can be influenced by cathepsin function. For example, altered PSAP or saposin function has been observed in Niemann‐Pick (SapA; CTSD, B, L, and S), Metachromatic Leukodystrophy (SapB; CTSD and L), and Gaucher's disease (SapC; CTSD, B, K, and S) (Amritraj et al. 2013; Gabande‐Rodriguez et al. 2019; Chung et al. 2016; Cermak et al. 2016; Seo et al. 2016; Alam et al. 2014; von Figura et al. 1983; von Bulow et al. 2002; Afinogenova et al. 2019; Tatti et al. 2012; Moran et al. 2000). Furthermore, loss of CTSD or CTSF activity directly causes neuronal ceroid lipofuscinoses (CLN10 and CLN13), a related group of lysosomal storage disorders characterized by lipofuscin accumulation and disrupted sphingolipid metabolism (Zhang et al. 2025). Thus, variations in cathepsin expression or activity may fine‐tune PSAP cleavage and thereby influence sphingolipid metabolism and disease progression.

Our lab and others have previously shown that CTSD activity is directly regulated by PGRN and MGFs (Butler, Cortopassi, Argouarch, et al. 2019; Butler, Cortopassi, Gururaj, et al. 2019; Zhou, Paushter, et al. 2017). Here, we showed that PGRN and MGF regulation of CTSD activity directly affects the rate of saposin production. First, we found that full‐length PGRN does not increase PSAP processing in vitro but rather seems to slow saposin production. While this finding was unexpected, it may be due to steric hindrance or structural alterations in the proteins. On the other hand, while the MGFs containing pG and BAC have small to negligible effects on the rate of saposin production, the MGF containing CDE increases the rate of saposin production by CTSD considerably, allowing it to produce saposins after only 5 min. The GrnD domain of PGRN has been reported to directly bind the intersaposin region between SapB and SapC in PSAP (Zhou, Sullivan, et al. 2017; Zhou, Sun, et al. 2017), suggesting that the direct binding of CDE to PSAP may be involved in the increased cleavage. We found that CTSD, despite being the most well‐studied cathepsin to process PSAP, is among the least efficient. Thus, CDE's considerable acceleration of its cleavage may have substantial effects in vivo both through depleting available PSAP and rapidly increasing abundance of saposins. Our study is the first to report direct influence of MGFs on the cleavage of PSAP, providing a biochemical mechanism that supports previous studies reporting that PGRN expression affects saposin abundance in cells (Zhou, Sun, et al. 2017; Valdez et al. 2020) and disrupting PGRN function leads to disease‐related dysfunction in lipid homeostasis (Evers et al. 2017; Huang et al. 2020; Arrant et al. 2019). Future studies that characterize the binding interaction of CTSD, PGRN/MGFs, and PSAP could provide further insight to the importance of this regulation in vivo.

Several studies indicate that PSAP cleavage and saposin generation vary with cell type and differentiation state. Classical biochemical work demonstrated cell‐specific processing patterns of PSAP (Leonova et al. 1996) and more recent studies report differential PSAP expression and cleavage in neural stem/progenitor populations and dopaminergic neurons (Kim et al. 2022; Labusch et al. 2024; He et al. 2023). These differences likely reflect both the protease expression landscape and the local lipid or proteolytic environment that modulate which cathepsins drive saposin production in vivo. For example, Kim et al. showed that ceramide‐dependent activation of cathepsin B promotes prosaposin cleavage into SapC, demonstrating how context‐specific lipid environments can steer which protease produces particular saposins (Kim et al. 2022). These cell‐type effects are further supported by single‐cell and atlas data showing that individual cathepsins have distinct expression patterns across tissues and cell types, implying that the predominant PSAP processing enzymes will vary by cell context (Human Protein Atlas).

In summary, our findings indicate that the processing of PSAP into saposins is subject to intricate regulatory mechanisms. These mechanisms involve protease‐specific cleavage sites within the intersaposin regions and different cleavage rates based on pH environments. Notably, our study reveals a novel role for the PSAP binding partner PGRN in directly regulating saposin production. Future investigations of the physiological significance of PSAP cleavage regulation by these proteases in vivo, particularly within the context of their cell‐type specific expression, will be of particular interest. Considering the key role of PSAP and saposins in sphingolipid metabolism and the existence of genetic mutations in PSAP associated with neurodegenerative diseases in humans (Park et al. 2020; Di Pardo and Maglione 2018; Johnson and Stolzing 2019; Castellanos et al. 2021; Schulze and Sandhoff 2014; Hernandez‐Diaz and Soukup 2020), it is clear that maintaining optimal levels of PSAP and saposins is crucial to maintain proper lysosomal function and sphingolipid homeostasis. With this study, we have identified PGRN, MGFs, and multiple cathepsins as key contributors to the production of saposins, which could present novel opportunities to modulate saposin levels in disease.

Author Contributions

Molly Hodul: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, funding acquisition, supervision. Courtney Lane‐Donovan: investigation, writing – review and editing, methodology, formal analysis, supervision. Emily S. Cheang: investigation, formal analysis. Vienna Gao: investigation, formal analysis. Paul J. Sampognaro: investigation, formal analysis, supervision. Edwina A. Mambou: investigation, formal analysis. Zoe Yang: visualization. Aimee W. Kao: conceptualization, investigation, writing – review and editing, project administration, supervision, resources.

Funding

This work was supported by F32 AG079659 received by M.H., K08 AG083050, R25 NS070680, Weill Neuroscience Investigator Award, and the Creative Minds Care Award received by C.L‐D, K08 NS121519 received by P.J.S, and RF1 NS1274114 received by A.W.K. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: jnc70357‐sup‐0001‐DataS1.pdf.

Hodul, M. , Lane‐Donovan C., Cheang E. S., et al. 2026. “Prosaposin Is Cleaved Into Saposins by Multiple Cathepsins in a Progranulin‐Regulated Fashion.” Journal of Neurochemistry 170, no. 1: e70357. 10.1111/jnc.70357.

Data Availability Statement

All data is contained within the manuscript. A preprint of this article was posted on bioRxiv on 02/16/2024 (https://doi.org/10.1101/2024.02.15.580326).